Bar codes are used in a number of applications to uniquely identify an item or group of items. The bar code is typically a combination of black bars and white spaces representing the unique identification code and is affixed to the item in any of a number of ways (e.g., a label). The reading and decoding of a bar code can be used to yield additional information concerning the item, such as a description of the item and its price. The bar code has also been used for inventory control, manufacturing parts control, and employee identification. Failure to properly read the bar code usually requires manual intervention, such as manual entry of the data (e.g., the identification number represented by the bar code). Manual intervention, however, can cause problems because of human error as well as causing delays and increasing costs.
Bar code scanners are the devices used to read the bar code affixed to an item so the desired information can be retrieved. In the case of a supermarket, one example of this would be the price associated with goods. The scanner reads the bar code by detecting the light reflected from the bar code and determining the light contrasts between the bars, typically the dark or black areas, and the spaces, typically the light or white areas. In most scanners the code is illuminated by a beam of coherent light (i.e., a laser) that moves or sweeps across the bar code. A rotating wheel with a number of mirror facets is typically used to sweep the coherent light across the bar code.
There is shown in FIG. 1, a typical omni-directional overhead scanner 10 for detecting a bar code 22 affixed to package 12 traveling on a conveyer belt 14 in direction 20. The scanner 10 illuminates the package top surface, where the bar code 22 is located, with pattern 16 of coherent light. In this way, the scanner 10 can read a bar code 22 that approaches the scanner 10 in any orientation. In this illustration, the bar code 22 is read so the package 12 can be sorted and sent along different paths 18a-c.
For one scanning technique, the bar code is read by sweeping a beam of coherent light across the entire bar code. In actual practice redundant passes or sweeps are made across the entire bar code to assure a good reading of the bar code, however, no reconstruction of the bar code is done. The redundant scanning is done to address possible localized width inaccuracies (e.g., local void in bar). This method is best suited for reading bar codes which have been pre-positioned so the bar code extends along the scan line of the sweeping coherent line. This bar code orientation has also been referred to as a "picket fence" or "ladder" orientation.
However, it is not always possible to control the orientation of the bar code with respect to the scan line of the scanning device. There are two ways that the traditional method could be used to scan a bar code which is at an angle with respect to the coherent light scan line. The bar code may be scanned with many beams of coherent light at numerous angles with respect to the surface of the bar code. This approach, however, requires a great deal of expensive equipment to implement. Alternatively, the bar codes may be made very tall so practically speaking the scan line will see the entire bar code. This approach significantly increases the size of bar code (e.g., the bar code label).
Another technique to read a bar code, which is at an angle with respect to a coherent light scan line, involves reconstructing the entire bar code from fragments of the bar code. This technique is based on the principle that, as the bar code moves under and through the coherent light scan line, the entire bar code will eventually be seen by the scanner. The scanner, however, only scans a portion or fragment of the bar code at a time, saving each of the scanned fragments. After the bar code has passed through the scan line, the complete bar code is reconstructed by reconnecting the saved scanned fragments.
When reconstructing a bar code, there is an increased risk that the bar code will be mis-read because of the loss in redundancy. Redundancy is the repetitive measuring of the same bar code element (e.g., same bar) to assure that the correct element width is provided for decoding. For any given scan, it is possible for an incorrect width to be outputted because of printing problems (e.g., voids or blotches), errors in the scanning process, and/or from wear and tear on the bar code (e.g., bar code label).
Since repetitive scanning of the entire bar code is not possible for the reconstruction technique, redundancy is obtained at the individual bar and space level by utilizing width data from overlapping regions of code fragments. When fragments are properly aligned, redundant information is available to refine and correct width data for the bars and spaces in the overlap regions. For reconstructive methods, therefore, redundancy is dependent upon getting a correct fragment alignment.
One type of reconstructive technique involves aligning fragments by locating and matching a bar and space pattern, including the bar and space widths, that is common to the fragments. This specific technique will be referred to hereinafter has as "pattern matching." One common cause for fragment misalignment with the pattern matching technique is the presence of repetitive patterns of bar elements in a bar code. Because of repetitive patterns there can be a number of areas in the fragments which are common to each other. Thus, while the bar and space patterns of the fragments may match, the fragments may not be properly aligned because the wrong bar and space patterns are being used for alignment.
Misalignment can occur for a number of other reasons. Human readable printing about the bar code can produce false bars at one or both ends. Printing errors as well as wear and tear on the bar code label can cause a particular bar to appear much smaller or larger in one scan than it does in another.
If exact pattern matches are required, very few bar codes will be successfully rebuilt, and the read rate will be very low. Since an exact pattern match will not result most of the time, an arbitrary limit is typically established so that fragments are considered in alignment if at least some set number of bars and spaces match. However, if the limit is set too loose, while the number of matches will go up and the read rate will be high, the number of mis-reads will go up. In practice, a majority of the mis-reads can be filtered out by using fixed length bar codes and checksums.
Therefore, determining accurate bar and space widths for proper decoding depends on a proper alignment of the bar code fragment overlapping regions. However, the proper alignment of the overlapping regions in known methods depends upon matching the patterns of the fragments. This method assumes that the bars and spaces have the correct widths. In sum, reconstructing bar codes by using known methods for pattern matching balances the need for accurate bar and space widths for decoding with the need for a reasonable read rate.
Reconstructive bar code scanning devices and the associated methodology are disclosed in U.S. Pat. Nos. 4,289,957, 4,488,678, 4,717,818, 4,973,829, 5,028,772, and 5,124,538. The differences between the above referenced patents is the specific manner in which the fragments are combined to reconstruct a complete bar code for decoding.
In U.S. Pat. Nos. 4,289,957, and 4,717,818 the bar code contains markers representing the ends and middle of the bar code. The scanning device reconstructs the complete bar code based on the presence or existence of markers in the scanned segments. In U.S. Pat. No. 4,289,957, a bar fragment portion not located between two markers (e.g., end and middle) is ignored and the fragment portions that lie between the markers are combined.
In U.S. Pat. No. 4,488,678, the superfluous or duplicative data from the overlap between the bar code fragments is removed in one of two ways. If a bar code has separation bars, the data outside the separation bars is eliminated. If the bar code does not have separation bars, the number of bars indicated by the combined data is compared to the known number of bars for the particular bar code being scanned. If the number of bars for the combined data is greater than the known value, the superfluous or duplicative data is removed from either of the overlapping regions.
In U.S. Pat. Nos. 4,973,829, 5,028,772, and 5,124,538, a complete bar code is reconstructed by relatively shifting the stored data of two fragments until there is a pattern matching of the data (i.e., the bar/space patterns match). The data from the fragments is then combined based on this point of commonalty. The differences between these patents is the manner in which the width data is relatively shifted, the amount and type of matching required, and how the data is reconstructed.
In U.S. Pat. No. 4,973,829, the data from multiple passes is combined using a superimposition technique. In this method, the data in the master memory is shifted one address at a time until the master memory data fully coincides with the data from one of the passes stored in the transaction memory. The data from the transaction memory is then superimposed on the data in the master memory starting at the point of coincidence.
In U.S. Pat. No. 5,028,772, a bar code is scanned producing two incomplete bar code segments where one fragment provides the beginning of the code, the other fragment provides the end, and both provide an overlapping middle portion. The two bar code fragments are combined by relatively shifting the overlapping regions of the fragments until a registered or pattern matched middle portion results. Essentially, the device slides one fragment along the data previously acquired until a pattern of bars and spaces match up.
U.S. Pat. No. 5,124,538 (a continuation of '772) describes a methodology whereby a plurality of scanned fragments can be combined to reconstruct the entire bar code. In this technique, while taking data relating to bar and space widths, data is also accumulated concerning the position of certain data events with respect to the starting point of each scan. The transition (e.g., white to black) position count for a valid middle portion of the first fragment is recorded and a range (+/-) is calculated. The second fragment is then analyzed to see if at least a portion of that fragment falls within the calculated range of the valid middle region of the first fragment. If it does not lie within the valid range, a fragment from a following scan is analyzed.
If the second fragment does fall within the calculated range, the valid middle region of the first fragment, the second fragment is shifted to each possible location within the calculated range. A pattern comparison is then made between the width data for each fragment at each possible location. Preferably, a bar/space pair is created by summing adjacent bars and spaces which are then used for pattern comparison purposes. If the pattern matches, then a registered middle portion of adjacent code fragments is identified or found.
Therefore, it is object of the present invention to provide a scanning device that avoids uncertainties inherent in the pattern matching alignment technique.
Another object of the present invention is to provide a scanning device that is more efficient and faster than prior art techniques.
It is a further object of the present invention to provide a scanning device that aligns scanned segments so that redundant information for each bar element scanned can be used to determine bar and space widths for decoding.
It is yet another object of the present invention to provide a scanning device that can reconstruct a complete bar code from scanned segments by calculating the positional change of a selected bar between scans.
It is yet a further object of the present invention to provide a scanning device that can align overlapping scanned code segments using the positional information of the selected bar.
It is still yet another object of the present invention to provide a scanning device that can be used for a number of applications including reading packages moving along high speed conveyers for data collection and/or sorting purposes.